Premium
More Sites or More Years? Optimal Study Design for Monitoring Fish Response to Watershed Restoration
Author(s) -
Liermann Martin,
Roni Phil
Publication year - 2008
Publication title -
north american journal of fisheries management
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.587
H-Index - 72
eISSN - 1548-8675
pISSN - 0275-5947
DOI - 10.1577/m06-175.1
Subject(s) - watershed , oncorhynchus , replicate , scale (ratio) , streams , abundance (ecology) , stream restoration , environmental science , fish <actinopterygii> , habitat , population , restoration ecology , fish migration , fishery , environmental resource management , dam removal , ecology , geography , computer science , cartography , statistics , biology , sediment , mathematics , machine learning , computer network , demography , sociology , paleontology
Every year in the Pacific Northwest, hundreds of stream restoration projects are implemented at great expense in the hope that they will increase salmonid abundance. Our understanding of how salmonids interact with their freshwater habitat has steadily improved, but we are still a long way from being able to reliably predict population‐level effects of individual projects. To determine whether these projects are in fact increasing salmonid abundance, we will need to implement restoration at the watershed scale, monitor the populations after the freshwater portion of their life history is completed, and replicate the experiment across multiple watersheds to produce results that can be generalized. Although there has been some progress in this direction, it has largely consisted of independent efforts at a relatively small scale. In this paper, we use smolt counts of coho salmon Oncorhynchus kisutch from streams in western Washington and Oregon, along with approximate restoration and monitoring costs, to estimate the most cost‐effective way of allocating monitoring effort between years and watersheds to allow detection of an average response to watershed‐scale restoration. We show that it is generally preferable to spread the available effort across more watershed pairs, unless the one‐time cost of adding a watershed pair is very high (e.g., when the cost of restoration is included). These results are sensitive to plausible changes in temporal and spatial variability and may change with different assumptions about response type, but they are derived using a logical, transparent process that incorporates available information.